From 2be6b143289e5479cc4a2667bb84e879116c2447 Mon Sep 17 00:00:00 2001 From: pennae Date: Mon, 29 Jan 2024 06:19:23 +0100 Subject: [PATCH] match line endings used by parser and error reports the parser treats a plain \r as a newline, error reports do not. this can lead to interesting divergences if anything makes use of this feature, with error reports pointing to wrong locations in the input (or even outside the input altogether). --- src/libutil/position.cc | 55 +++++++++++-------- src/libutil/position.hh | 42 ++++++++++++++ tests/functional/lang/eval-fail-eol-1.err.exp | 6 ++ tests/functional/lang/eval-fail-eol-1.nix | 3 + tests/functional/lang/eval-fail-eol-2.err.exp | 6 ++ tests/functional/lang/eval-fail-eol-2.nix | 2 + tests/functional/lang/eval-fail-eol-3.err.exp | 6 ++ tests/functional/lang/eval-fail-eol-3.nix | 3 + 8 files changed, 99 insertions(+), 24 deletions(-) create mode 100644 tests/functional/lang/eval-fail-eol-1.err.exp create mode 100644 tests/functional/lang/eval-fail-eol-1.nix create mode 100644 tests/functional/lang/eval-fail-eol-2.err.exp create mode 100644 tests/functional/lang/eval-fail-eol-2.nix create mode 100644 tests/functional/lang/eval-fail-eol-3.err.exp create mode 100644 tests/functional/lang/eval-fail-eol-3.nix diff --git a/src/libutil/position.cc b/src/libutil/position.cc index b39a5a1d4..724e560b7 100644 --- a/src/libutil/position.cc +++ b/src/libutil/position.cc @@ -29,32 +29,17 @@ std::optional Pos::getCodeLines() const return std::nullopt; if (auto source = getSource()) { - - std::istringstream iss(*source); - // count the newlines. - int count = 0; - std::string curLine; - int pl = line - 1; - + LinesIterator lines(*source), end; LinesOfCode loc; - do { - std::getline(iss, curLine); - ++count; - if (count < pl) - ; - else if (count == pl) { - loc.prevLineOfCode = curLine; - } else if (count == pl + 1) { - loc.errLineOfCode = curLine; - } else if (count == pl + 2) { - loc.nextLineOfCode = curLine; - break; - } - - if (!iss.good()) - break; - } while (true); + if (line > 1) + std::advance(lines, line - 2); + if (lines != end && line > 1) + loc.prevLineOfCode = *lines++; + if (lines != end) + loc.errLineOfCode = *lines++; + if (lines != end) + loc.nextLineOfCode = *lines++; return loc; } @@ -109,4 +94,26 @@ std::ostream & operator<<(std::ostream & str, const Pos & pos) return str; } +void Pos::LinesIterator::bump(bool atFirst) +{ + if (!atFirst) { + pastEnd = input.empty(); + if (!input.empty() && input[0] == '\r') + input.remove_prefix(1); + if (!input.empty() && input[0] == '\n') + input.remove_prefix(1); + } + + // nix line endings are not only \n as eg std::getline assumes, but also + // \r\n **and \r alone**. not treating them all the same causes error + // reports to not match with line numbers as the parser expects them. + auto eol = input.find_first_of("\r\n"); + + if (eol > input.size()) + eol = input.size(); + + curLine = input.substr(0, eol); + input.remove_prefix(eol); +} + } diff --git a/src/libutil/position.hh b/src/libutil/position.hh index a184997ed..9bdf3b4b5 100644 --- a/src/libutil/position.hh +++ b/src/libutil/position.hh @@ -67,6 +67,48 @@ struct Pos bool operator==(const Pos & rhs) const = default; bool operator!=(const Pos & rhs) const = default; bool operator<(const Pos & rhs) const; + + struct LinesIterator { + using difference_type = size_t; + using value_type = std::string_view; + using reference = const std::string_view &; + using pointer = const std::string_view *; + using iterator_category = std::input_iterator_tag; + + LinesIterator(): pastEnd(true) {} + explicit LinesIterator(std::string_view input): input(input), pastEnd(input.empty()) { + if (!pastEnd) + bump(true); + } + + LinesIterator & operator++() { + bump(false); + return *this; + } + LinesIterator operator++(int) { + auto result = *this; + ++*this; + return result; + } + + reference operator*() const { return curLine; } + pointer operator->() const { return &curLine; } + + bool operator!=(const LinesIterator & other) const { + return !(*this == other); + } + bool operator==(const LinesIterator & other) const { + return (pastEnd && other.pastEnd) + || (std::forward_as_tuple(input.size(), input.data()) + == std::forward_as_tuple(other.input.size(), other.input.data())); + } + + private: + std::string_view input, curLine; + bool pastEnd = false; + + void bump(bool atFirst); + }; }; std::ostream & operator<<(std::ostream & str, const Pos & pos); diff --git a/tests/functional/lang/eval-fail-eol-1.err.exp b/tests/functional/lang/eval-fail-eol-1.err.exp new file mode 100644 index 000000000..3f5a5c22c --- /dev/null +++ b/tests/functional/lang/eval-fail-eol-1.err.exp @@ -0,0 +1,6 @@ +error: undefined variable 'invalid' + at /pwd/lang/eval-fail-eol-1.nix:2:1: + 1| # foo + 2| invalid + | ^ + 3| # bar diff --git a/tests/functional/lang/eval-fail-eol-1.nix b/tests/functional/lang/eval-fail-eol-1.nix new file mode 100644 index 000000000..476223919 --- /dev/null +++ b/tests/functional/lang/eval-fail-eol-1.nix @@ -0,0 +1,3 @@ +# foo +invalid +# bar diff --git a/tests/functional/lang/eval-fail-eol-2.err.exp b/tests/functional/lang/eval-fail-eol-2.err.exp new file mode 100644 index 000000000..ff13e2d55 --- /dev/null +++ b/tests/functional/lang/eval-fail-eol-2.err.exp @@ -0,0 +1,6 @@ +error: undefined variable 'invalid' + at /pwd/lang/eval-fail-eol-2.nix:2:1: + 1| # foo + 2| invalid + | ^ + 3| # bar diff --git a/tests/functional/lang/eval-fail-eol-2.nix b/tests/functional/lang/eval-fail-eol-2.nix new file mode 100644 index 000000000..0cf92a425 --- /dev/null +++ b/tests/functional/lang/eval-fail-eol-2.nix @@ -0,0 +1,2 @@ +# foo invalid +# bar diff --git a/tests/functional/lang/eval-fail-eol-3.err.exp b/tests/functional/lang/eval-fail-eol-3.err.exp new file mode 100644 index 000000000..ada3c5ecd --- /dev/null +++ b/tests/functional/lang/eval-fail-eol-3.err.exp @@ -0,0 +1,6 @@ +error: undefined variable 'invalid' + at /pwd/lang/eval-fail-eol-3.nix:2:1: + 1| # foo + 2| invalid + | ^ + 3| # bar diff --git a/tests/functional/lang/eval-fail-eol-3.nix b/tests/functional/lang/eval-fail-eol-3.nix new file mode 100644 index 000000000..33422452d --- /dev/null +++ b/tests/functional/lang/eval-fail-eol-3.nix @@ -0,0 +1,3 @@ +# foo +invalid +# bar